|
BobHoward posted:so far it's been very niche since performance in existing non-apple arm SoCs just isn't competitive with x86. qualcomm's looking to change that I'm kinda hoping they do, but not exactly hopeful from the rumors I've read about their up and coming arm chip
|
# ? Mar 8, 2024 14:45 |
|
|
# ? May 16, 2024 17:36 |
|
champagne posting posted:I'm kinda hoping they do, but not exactly hopeful from the rumors I've read about their up and coming arm chip spill!
|
# ? Mar 8, 2024 14:51 |
|
the sole real-world use case for windows arm64 is running windows vms on macs, and the x86 emulation works pretty well there ime
|
# ? Mar 8, 2024 23:52 |
|
champagne posting posted:I'm kinda hoping they do, but not exactly hopeful from the rumors I've read about their up and coming arm chip I got my arm chips boosted at the same time I got my flu shot
|
# ? Mar 9, 2024 16:21 |
|
still slowly refining lil i added conforming semantics to dictionaries, so i can do things like sum/union a pair of dicts without explicit loops: generalized amending assignments to work with tables (copy-on-write; tables themselves are immutable values), which is much more straightforward than update for some tasks, especially if the target column varies: and introduced a like operator for glob matching, which makes it way easier to do 1:1 translations of a lot of sql examples. i've also found this surprisingly handy in practical decker scripts, since it's now much easier to do things like query for all the widgets on a card whose name has a particular suffix or add fuzzy search to crud programs i also just finished writing Learn Lil in 10 Minutes, which is a nice fast-paced language overview for folks who already know how to program
|
# ? Mar 21, 2024 04:43 |
|
nifty language! sql stuff on tables is . what's the motivation behind the right-to-left precedence? i feel like that'd be hard for me to remember heh.
|
# ? Mar 21, 2024 13:30 |
|
uniform right-to-left precedence is common to Q, J, K, APL, and other languages in the same family (Lil isn't a "pure" APL-descendant, but it does borrow many ideas from K and Q) with a large number of primitive operators, i think uniform precedence is much easier to remember than a complex tower that tries to "do the right thing" for various operators, and it simplifies the implementation slightly certainly takes some getting used to at first
|
# ? Mar 21, 2024 15:11 |
|
one can argue a lot about it as a policy, but it is certainly not a "hard to remember", in that the thing to remember begins and ends with the phrase "everything is right-to-left" also i'd argue that it is simply better as honestly you should probably parenthesize pretty much everything beyond mixing multiplication and addition in any language with mixed associativity/precedence
|
# ? Mar 21, 2024 15:22 |
|
ehh i assumed it would trip me up for a bit but yeah, i don't think it's bad or anything. cool honestly. i did some K maybe 15 years ago for some challenge and while i don't remember that rule, i finished the challenge so it clearly wasn't that difficult to remember.
defmacro fucked around with this message at 15:35 on Mar 21, 2024 |
# ? Mar 21, 2024 15:32 |
|
ultimately the main thing that i loved about k/kdb/q was first class tables as a data structure. in my experience it did *a lot* for program structure in a dynamically typed language. you just naturally wind up writing all interfaces in terms of tables, including then proper names for things and a typespec. a table is just a pretty human-friendly format. while also being pretty much gold standard for things to write out to storage. afaik ij's lil is quite unique among the open source descendants to pick that part up.
|
# ? Mar 21, 2024 17:57 |
|
it took me a long time to appreciate the value of tables in k and q; having them as a first-class datatype really helps simplify lots of things. i've lost count how many times i've written programs in javascript that operated on lists of dictionaries that all have the same keyset by convention, but with a table you can express the same thing and still capture meaningful information in the 0-record case i think the rise in popularity of dataframes (usually as second-class entities) reflects a strong desire for tabular structures and operations for manipulating them uniformly; perhaps in the coming decades this sort of feature will gain popularity in the same way that many mainstream languages have adopted functional programming features
|
# ? Mar 21, 2024 18:19 |
|
given that expressions evaluate right-to-left, I found it weird that dict had the keys on the left, because for some reason it seems like keys should be before values in a definition like that I don’t relish having to maintain a large dictionary with the keys and values distant from each other, but you can probably transform a list of pairs into a dictionary with 3 characters (none alphanumeric) once you know what you’re doing
|
# ? Mar 21, 2024 20:24 |
|
Subjunctive posted:given that expressions evaluate right-to-left, I found it weird that dict had the keys on the left, because for some reason it seems like keys should be before values in a definition like that 2
|
# ? Mar 21, 2024 20:25 |
|
nah, but kind of agreed that the map literal syntax is dubious. it makes sense in that (k/q) very much treats it as a pair of lists in a bunch of ways, and it makes the parse easy, but i can't make a similar case that it is human-friendly.
|
# ? Mar 21, 2024 20:28 |
|
dict is just an operator for creating dictionaries from lists; it's not a literal syntax and certainly not the only way to obtain a dict i do agree that there are a few situations where flipping the arguments might be more convenient, but i wasn't able to convince myself it was a slam-dunk win one of the lowest-friction ways of making dictionaries (or nested dictionaries) is to use the "amending" assignment syntax to construct them implicitly code:
|
# ? Mar 21, 2024 21:04 |
|
Cybernetic Vermin posted:ultimately the main thing that i loved about k/kdb/q was first class tables as a data structure. in my experience it did *a lot* for program structure in a dynamically typed language. you just naturally wind up writing all interfaces in terms of tables, including then proper names for things and a typespec. the children yearn for
|
# ? Mar 22, 2024 06:32 |
|
redleader posted:the children yearn for sounds more like mainframes
|
# ? Mar 22, 2024 10:22 |
|
NihilCredo posted:sounds more like mainframes and what is a mainframe but a proto-excel?
|
# ? Mar 22, 2024 11:13 |
|
you're the mainframe now, dec
|
# ? Mar 22, 2024 12:00 |
|
Is anyone here familiar with SkookumScript? It was a scripting language for the Unreal 4 engine, and is mostly dead/unmaintained now, but it seems very interesting in terms of syntax/semantics. I think the folks behind it got hired by Epic and are working (with others) on that Verse language.
|
# ? Mar 25, 2024 22:27 |
|
sounds like a real choocher
|
# ? Mar 25, 2024 22:50 |
|
ColTim posted:Is anyone here familiar with SkookumScript? It was a scripting language for the Unreal 4 engine, and is mostly dead/unmaintained now, but it seems very interesting in terms of syntax/semantics. I think the folks behind it got hired by Epic and are working (with others) on that Verse language. that was a fun read, thank you!
|
# ? Mar 25, 2024 22:58 |
|
the new language has documentation already: https://dev.epicgames.com/documentation/en-us/uefn/verse-language-reference but you can, uh, only use it in fortnite. that's not a euphemism or something, it only works in the unreal editor for fortnite right now, and may never come to the "real" unreal engine
|
# ? Mar 25, 2024 23:11 |
|
abraham linksys posted:the new language has documentation already: https://dev.epicgames.com/documentation/en-us/uefn/verse-language-reference oh BOO
|
# ? Mar 26, 2024 00:15 |
|
a programming language that has absolutely no standalone implementation outside of a specific entire AAA video game is inexcusable how does a company worth tens of billions of dollars owned by a longstanding enthusiastic plt weenie screw this up
|
# ? Mar 26, 2024 03:03 |
|
what? completely proprietary programming languages are the best kind of vendor lock-in, this is why companies are worth tends of billion dollars in the first place
|
# ? Mar 26, 2024 03:09 |
|
I almost went to work on that last year, but then it evaporated... and then epic games had their huge layoffs. Oh. That's why. But I imagine it's a bit of conflicting priorities thing happening. They knew creating a new languages was going to be an long and expensive lift. And they committed to it. And then... well, expensive suddenly became more expensive (hence layoffs) and "metaverse" stopped being the hot new investor thing. I imagine there's a lot of sad trade-offs happening over there.
|
# ? Mar 26, 2024 03:16 |
|
waah doing math on fractions in latex is so compliated
|
# ? Mar 26, 2024 13:51 |
|
I would simply not do math on fractions in a typesetting system
|
# ? Mar 26, 2024 13:56 |
|
One of the darkest events in computer history was when Guy Steele convinced Knuth to turn TeX into a programming language.
|
# ? Mar 26, 2024 14:16 |
|
Bloody posted:I would simply not do math on fractions in a typesetting system tbh my brain is correctly broken to program latex now, so the hard part is actually doing the math, because im stupid.
|
# ? Mar 26, 2024 14:19 |
|
Athas posted:One of the darkest events in computer history was when Guy Steele convinced Knuth to turn TeX into a programming language. yeah, rare miss by both of them
|
# ? Mar 26, 2024 14:42 |
|
Subjunctive posted:yeah, rare miss by both of them especially since it happened *after* scheme was designed. i largely think some programming language was inevitable, it just didn't have to be as crude as tex is. but afaik there's no reason to use anything but luatex anymore so that's all in the past. Cybernetic Vermin fucked around with this message at 15:08 on Mar 26, 2024 |
# ? Mar 26, 2024 15:05 |
|
Bloody posted:I would simply not do math on fractions in a typesetting system have you tried doing math on a Linotype, or mayhaps a Monotype?
|
# ? Mar 26, 2024 16:00 |
|
Athas posted:One of the darkest events in computer history was when Guy Steele convinced Knuth to turn TeX into a programming language instead of just a set of Lisp macros and functions.
|
# ? Mar 26, 2024 19:35 |
|
pick any technology that doesn't directly involve lisp or sexprs if lisp weenies had their way when that technology was designed it would be infinitely worse than whatever we have today
|
# ? Mar 27, 2024 00:54 |
|
Internet Janitor posted:pick any technology that doesn't directly involve lisp or sexprs
|
# ? Mar 27, 2024 00:56 |
|
points at language with a repl: "lisp" points at language with automatic memory management: "lisp" (crying) "phil, you can't just point at things and say they contain an ad-hoc, incomplete common lisp implementation" points at seagull carrying a curved french fry: "lisp"
|
# ? Mar 27, 2024 01:04 |
|
Internet Janitor posted:points at language with a repl: "lisp" This is a molehill I will tie-dye on.
|
# ? Mar 27, 2024 02:05 |
|
|
# ? May 16, 2024 17:36 |
|
I just committed a change to my team's Java-only project that says how to do something using the Kotlin REPL and I keep laughing at how obnoxious that is I tried to do it in the Java REPL first, I swear! It wouldn't put the project classes in the classpath by default, though, and the Kotlin REPL somehow did.
|
# ? Mar 27, 2024 02:08 |